Are There Preferred Viewing Locations in Chinese Reading? Evidence from Eye-Tracking and Computer Simulations

The Chinese writing system is different from English in that individual words both comprise one to four characters and are not separated by clear word boundaries (e.g., interword spaces). These differences raise the question of how readers of Chinese know where to move their eyes to support efficient lexical processing? The widely accepted "default-targeting hypothesis" suggests that Chinese readers direct their eyes to a small number of "preferred-viewing locations" (PVLs), such as the beginning or middle of upcoming words. In this article, we report two eye-movement experiments testing this hypothesis. In both experiments, participants read sentences comprising entirely two-character words, but either without (Experiment 1) or with (Experiment 2) explicit knowledge of this structure prior to their participation. The results of both experiments indicate the absence of PVLs. Simulations using implemented versions of a simple oculomotor-based hypothesis, two variants of the default-targeting hypothesis, and the hypothesis that saccade lengths are modulated as a function of estimated parafoveal-processing difficulty (i.e., dynamic-adjustment hypothesis) suggest that the latter provides the best account of saccadic-targeting during Chinese reading. These results are discussed in relation to broader issues of eye-movement control during reading and how models of such must be modified to provide more accurate accounts of the reading of Chinese and other languages.